Numerical study of gas entrainment phenomena, II

Tanaka, Nobuatsu*; Ohshima, Hiroyuki

In this study, we carried out some numerical benchmark in order to survey effectiveness of some LES methods for the phenomena of gas entrainment. Considering the complicated geometry problems, we chose a collocated variable system combined with the high order scheme proposed by Inagaki. In numerical analysis of the Moriya's experiment, we focused on the effect of Smagorinsky model (SM) and Dynamic Smagorinsky Model (DSM) with LISEA (Local Inter-Scale Equilibrium Assumption of Energy), propsed by Murota, and in numerical analysis of the Okamoto's experiment, we focused on the effectiveness of LISEA.(1) Numerical analysis of Moriya's experimentWe compared the results obtained by the LISEA with those of the SM. In such a cylindrical geometry, the empirical parameter of SM is not known, so we performed two cases of simulations with Cs=0.1 and 0.2. Comparing the velocity profiles, clear difference between SM (Cs=0.1) and LISEA was not observed. In the case of Cs=0.2, the results are worse than the results of the other two cases. The eddy viscosity distributions of SM and LISEA are quite different. While the results by the SM are depends on the definition of empirical parameter, the LISEA do not need any empirical parameters or dumping function. Consequently, the LISEA is superior to the SM in the case of such a complicated problems. (2) Numerical analysis of Okamoto's experimentWe surveyed effects of turbulence models through comparison of numerical results of DNS and LISEA models. From viewpoint of some distinguishing physical quantities, there is little difference between the results of DNS and those of LISEA. Since the reversed flow around the outlet boundaries, which is not observed in the LISEA simulation, was observed in the DNS simulation, the LISEA makes an effect of numerical viscosity and stabilize the numerical solution.